Scaling RNA Production From The Lab To The Clinic

A conversation with John Stubenrauch and Sam Deutsch, Nutcracker Therapeutics

Ramping up RNA therapeutic production to clinical scale brings unique challenges compared to traditional biologics.

Nutcracker Therapeutics is preparing to enter the clinic with its lead candidate, NTX-250, a multimodal RNA therapeutic for HPV-derived tumors.

Central to the company’s strategy, the Nutcracker Manufacturing Unit platform, or NMU, uses biochips and microfluidics to standardize many of the RNA manufacturing variables. It’s also highly scalable.

While this might be a first for Nutcracker, its leaders have deep experience in the art and science of scaling.

We asked two of them, former Chief Operating Officer John Stubenrauch and Chief Scientific Officer Sam Deutsch, some wide-ranging questions about the scale-up process, including how it compares to traditional biologics, how they’re overcoming technology shortcomings, and how the industry wants to address purity challenges with synthetic templates. Soon after our conversation, Stubenrauch left Nutcracker to join Day One Biopharmaceuticals as chief technology officer.

Their answers have been edited for brevity and clarity. Here’s what they told us:

Let‘s go back and look at the standard process transfer from small- to large-scale production for traditional biologics. What are the challenges, and how do they compare to scaling for RNA?

Stubenrauch: When you look at the ability to scale up, we’re not talking about 2,000 liters to 20,000 liters, where there are complexities in mixing performance, wall effects, geometric designs, etc. For mRNA manufacturing, large scale typically means 10 liters to 30 liters, which enable production at the gram scale.

With our biochip based system, you can move from microgram to gram scales with very modest differences in chip design.

We’ve demonstrated very clearly that based on the proprietary design of our mixing chambers we are seeing very consistent performance across different scales with the basic footprint of the instrument staying essentially the same. Physically, you can see that there are differences in the chamber dimensions, but it’s not on the order of magnitude seen with traditional bioreactors.

As you’re moving from preclinical to clinical, what considerations are you making?

Stubenrauch: In the preclinical development phase, you’re doing initial characterization of your product. You’re still defining your critical quality attributes. As you start going to the clinical setting, you add to your characterization library.

You start looking at not just the properties of the product you’re making, but you’re also building your control strategy around that.

When you get to the clinical phase, now you’re starting to put more specifications in place and also in-process controls to demonstrate that the process is running in a more consistent manner. It’s not just the quality attributes, it’s also some of the control variables.

It’s a build. As you go from preclinical to clinical stage, it’s collecting more information, more process understanding, and putting more of your control variables in place.

Deutsch: I would say the one thing that is different with RNA therapeutics relative to other modalities is this concept of platform. The biophysical properties of RNA are quite similar regardless of the sequence that encodes one therapeutic versus another one. It means that a lot of your analytical methods and control strategies are quite transferable across different RNA drugs.

In practice, the platform concept means that once you’ve gone through the exercise of validating your assays, supply chain, and control strategy for one product, you have a pretty powerful toolset that you can bring to hopefully accelerate the next set of products coming down your pipeline.

It’s particularly relevant when you get to the personalized setting as well, where the RNA sequences will be different for every patient.

I’d add an important caveat: one area where things will change is on the formulation side. Different routes of administrations will result in formulation changes, which will typically require new data generation, for example, concerning safety and stability and fill/finish strategies.

Sam, you mentioned fill/finish. Can you tell us more about specific issues with that part of production?

Deutsch: The challenge gets more difficult as your batch volume gets smaller. For example, if you’re making a large batch, and it’s the material that you’re going to use for your whole Phase 1 or Phase 2 study, you solve the problem once. You run your set of QC (quality control) for release once, and you’re done.

In RNA, it can be far down on the other end of the spectrum where you’re making drugs for one or a few patients. They’re highly personalized. In that setting, most of the fill/finish technologies that have been developed for thousands of vials are not well-adapted for making 30 vials, which is what you might need for one patient.

The same is true for the QC methods — releasing manufacturing batches a few times per year is very different from having to do hundreds of lot releases every week, as one needs to do for RNA personalized cancer products. That’s where the thinking and the technology to address that becomes an interesting problem, and both the fill/finish and batch release strategies have to be re-imagined.

How are you addressing these particular issues at Nutcracker?

Stubenrauch: We’re always looking at what might be available. There are some technologies that address those issues. We’re also qualifying our equipment and testing it to make sure we understand its robustness.

I think with sterile filling, which I have a lot of experience with, even at large scale, it’s really about ensuring that you’re educating your staff on the principles of good aseptic filling. It starts with your incoming raw materials, your incoming product contact components, your reagents, etc. Making sure everything that comes in is clean, practicing good gowning procedures, and employing a good environmental monitoring program — sterility assurance in filling is about layers of protection to provide assurance that the end product is good.

When you’re doing multiple changeovers, it gives you practice, which is the good news. But your highest point of variability is during those changeovers. Once you have the filling operation running, it’s typically pretty stable.

Is Nutcracker using in-silico modeling and, if so, how have you deployed that to bridge data gaps when scaling?

Deutsch: One approach we have taken that I think is particularly important for RNA is the concept of designing for quality. Every protein sequence can be encoded as RNA by trillions of different sequences, so picking a good sequence that is likely to have the right potency profile and also yield the right quality attributes is critical. Predicting potential directly from sequences is a tractable problem providing the correct models are available .

So, taking the time to build great models is key — in our case, we built an artificial intelligence tool — with pretty high confidence to assign a factorability score to each design.

If you’re in a setting where you produce lots of small and different batches, every time you fail will have a significant impact on your operations as a whole because then you have to map your re-work activities. Re-work becomes very expensive and very difficult to track. The way we have tried to address this is to design carefully so that we know for 95% to 98% of the batches, you’re going to meet your specifications for lot release.

This concept of design for quality is something that John has been driving quite hard in the company — let’s do the best we can to harness the power of AI.

Stubenrauch: Sam’s team has done an incredible job here of contemplating manufacturability, design and sequence of products that we’re making. In many ways, it’s very predictable. It’s even predictable to the point of anticipating what the variability might be in yield — which constructs we might expect to be lower yield versus higher yield.

It allows us to pre-plan and use that knowledge base to build and improve manufacturing performance. So the quality-by-design approach is a key element to our success.

Speaking of quality, our next question is related to consistency in RNA size, integrity, and endotoxin levels during scale-up. Is that an issue where you may have inconsistencies?

Stubenrauch: I have not seen inconsistencies or any issues with the scale of operability. When you look at endotoxin controls (and controls in general), given the fact that we’re leveraging single-use technologies in addition to looking at our incoming materials,designing and understanding the properties early on in our process has made that transition much more straightforward.

We’ve run thousands of cycles on our manufacturing unit and, particularly for endotoxins, we have not seen a problem because of the controls and thoughtfulness that go into the product from the beginning.

Deutsch: One area that deserves special consideration concerns the source of DNA, which is a critical raw material in the RNA manufacturing process.

For large-scale RNA manufacturing, most people will source a bacterially-derived DNA template. Typically, the end result of a drug substance made with a bacterially-derived template is higher endotoxin levels; this is something one needs to keep an eye on.

The alternative is to work with synthetic templates that are not bacterially derived. We love when we can stick to our synthetic templates for as long as possible. There are solutions we’re exploring to do that even at larger scales – and others in the industry are exploring that, too. We’ve heard from other large-scale manufacturers that they are investing significant resources in making synthetic templates for production at liter scales. I would be very supportive of that structure.

Are there other benefits beyond endotoxin control that come with using synthetic templates?

Deutsch: From a process point of view, it’s cleaner; it’s more reproducible. There are certain sequences that bacteria don’t like, so when you change from plasmid to plasmid, you’ll get yield variation, which can become problematic.

Typically, on the synthetic side, things are pretty consistent and there are also opportunities to reduce the cycle time.

About The Experts:

John Stubenrauch is chief technology officer at Day One Biotherapeutics. He is the former chief operating officer at Nutcracker Therapeutics where he oversaw manufacturing, supply, quality, and product technical operations. Before Nutcracker, he was a vice president at Gilead Sciences following the acquisition of Immunomedics, where he was senior vice president of manufacturing. He has 25 years of experience in the biopharma industry supporting the manufacture of ADCs, biologics, vaccines, and synthetic peptides and molecules. Before Immunomedics, he worked at AstraZeneca and Merck. Stubenrauch earned a Ph.D. in chemical engineering from the University of Pennsylvania and an MBA from the Wharton School of Business.

Sam Deutsch is chief scientific officer at Nutcracker Therapeutics where he leads biological research and development. He orchestrates the advancements of Nutcracker’s therapeutics pipeline through close interactions between internal research and development, scientific advisors, and clinical collaborators. Prior to Nutcracker, Sam led the DNA synthesis platform at the Joint Genome Institute, part of Lawrence Berkeley National Laboratory. Sam’s experience also includes research in molecular hematology at the Geneva University Hospital and cancer genomics at the World Health Organization. He earned a B.Sc. in genetics from University of Nottingham and received his Ph.D. from the University of Geneva. He is the author of over 75 peer reviewed papers.